Mechanical and chemical-mechanical planarizing processes (collectively "CMP") are used in the manufacturing of microelectronic devices for forming a flat surface on semiconductor wafers, field emission displays and many other microelectronic substrates. FIG. 1 schematically illustrates a planarizing machine 10 with a platen or table 20, a carrier assembly 30 over the table 20, a polishing pad 40 on the table 20, and a planarizing fluid 44 on the polishing pad 40. The planarizing machine 10 may also have an under-pad 25 between the platen 20 and the polishing pad 40. In many planarizing machines, a drive assembly 26 rotates (arrow A) and/or reciprocates (arrow B) the platen 20 to move the polishing pad 40 during planarization.
The carrier assembly 30 controls and protects a substrate 12 during planarization. The carrier assembly 30 typically has a substrate holder 32 that holds the substrate 12 via suction, and a pad 34 in the substrate holder 32 that supports the backside of the substrate 12. A drive assembly 36 of the carrier assembly 30 typically rotates and/or translates the substrate holder 32 (arrows C.sub.1 and D, respectively). The substrate holder 32, however, may be a weighted, free-floating disk (not shown) that slides over the polishing pad 40.
The combination of the polishing pad 40 and the planarizing fluid 44 generally define a planarizing medium that mechanically and/or chemically-mechanically removes material from the surface of the substrate 12. The polishing pad 40 can be a conventional non-abrasive polishing pad without abrasive particles composed of a polymeric material (e.g., polyurethane), or it can be an abrasive polishing pad with abrasive particles fixedly bonded to a suspension material. In a typical application, the planarizing fluid 44 may be a CMP slurry with abrasive particles and chemicals for use with a conventional nonabrasive polishing pad. In other applications for use with an abrasive polishing pad, the planarizing fluid 44 is generally a "clean" chemical solution without abrasive particles.
To planarize the substrate 12 with the planarizing machine 10, the carrier assembly 30 presses the substrate 12 against a planarizing surface 42 of the polishing pad 40 in the presence of the planarizing fluid 44 (arrow C.sub.2). The platen 20 and/or the substrate holder 32 then move relative to one another to translate the substrate 12 across the planarizing surface 42. As a result, the abrasive particles and/or the chemicals in the planarizing medium remove material from the surface of the substrate 12.
CMP processes should consistently and accurately produce a uniformly planar surface on the substrate assembly to enable precise fabrication of circuits and photo-patterns. During the fabrication of transistors, contacts, interconnects, and other components, many substrate assemblies develop large "step heights" that create a highly topographic substrate surface. To enable the fabrication of integrated circuits with high densities of components, it is necessary to produce a planar substrate surface at several stages of processing the substrate assembly because non-planar substrate surfaces significantly increase the difficulty of forming sub-micron features or photo-patterns to within a tolerance of approximately 0.1 .mu.m. Thus, CMP processes should typically transform a highly topographical substrate surface into a highly uniform, planar substrate surface (e.g., a "blanket surface").
In the competitive semiconductor industry, it is also highly desirable to maximize the yield of operable devices as quickly as possible. One factor of CMP processing that affects the yield of operable devices is the ability to accurately stop CMP processing at a desired endpoint. In a typical CMP process, the desired endpoint is reached when the surface of the substrate is highly planar and/or when enough material has been removed from the substrate assembly to form discrete components of the integrated circuits (e.g., shallow-trench-isolation structures, contacts, damascene lines, etc.). Accurately endpointing CMP processing is important for maintaining a high yield because: (1) subsequent processing may not be possible if the surface is not sufficiently planar; and/or (2) the integrated circuits may not operate if the discrete components are not accurately formed. For example, if the substrate is "under-planarized," shallow-trench-isolation structures may not be adequately isolated from one another. Conversely, if the substrate assembly is "over-polished," "dishing" can occur in shallow-trench-isolation structures that can cause current-leakage paths or parasitic capacitance. Extreme cases of over-polishing can even destroy sections of the substrate assembly. Thus, it is highly desirable to stop CMP processing at the desired endpoint.
One drawback of CMP processing is that it is difficult to determine when the substrate surface is both planar and at the desired endpoint elevation in the substrate assembly. In one conventional method for determining the endpoint of CMP processing, the planarizing period of one substrate assembly in a run is estimated using the polishing rate of previous substrate assemblies in the run and the thickness of material that is to be removed from the particular substrate assembly. The estimated planarizing period for the particular substrate assembly, however, may not be accurate because the polishing rate may change from one substrate assembly to another. Thus, this method may not accurately planarize all of the substrate assemblies in a run to the desired endpoint.
In another method for determining the endpoint of CMP processing, the substrate assembly is removed from the pad and the substrate carrier, and then a measuring device measures a change in thickness of the substrate assembly. Removing the substrate assembly from the pad and substrate carrier, however, is time-consuming and may damage the substrate assembly. Thus, this method generally reduces the throughput and yield of CMP processing.
In still another method for determining the endpoint of CMP processing, a portion of the substrate assembly is moved beyond the edge of the pad, and an interferometer directs a beam of light directly onto the exposed portion of the substrate assembly to measure a change in thickness of a transparent layer. The substrate assembly, however, may not be in the same reference position each time it overhangs the pad. For example, because the edge of the pad is compressible, the substrate assembly may not be at the same elevation for each measurement. Thus, this method may inaccurately measure the change in thickness of the substrate assembly.
In yet another method for determining the endpoint of CMP processing, U.S. Pat. Nos. 5,036,015 and 5,069,002, which are herein incorporated by reference, disclose detecting the planar endpoint by sensing a change in friction between a wafer and the polishing medium. Such a change of friction may be produced by a different coefficient of friction at the wafer surface as one material (e.g., an oxide) is removed from the wafer to expose another material (e.g., a metal film). More specifically, U.S. Pat. Nos. 5,036,015 and 5,069,002 disclose detecting the change in friction by measuring the change in electrical current through the drive motor for the platen and/or substrate holder.
Although the endpoint detection technique disclosed in U.S. Pat. Nos. 5,036,015 and 5,069,002 is an improvement over the previous endpointing methods, the increase in current through the drive motors may not accurately indicate the endpoint of a substrate. The detection of a single change in friction at the interface between the different materials may only indicate that at least a portion of the substrate surface is at the level of the interface. Other portions of the substrate surface, however, may be above or below the interface level and/or the interface level itself may not be planar. The apparatus and methods disclosed in U.S. Pat. Nos. 5,036,015 and 5,069,002 may accordingly indicate that at least a portion of the substrate surface is at the endpoint elevation, but they do not necessarily indicate that the substrate surface is planar. Thus, the apparatus and methods of U.S. Pat. Nos. 5,036,015 and 5,069,002 may not indicate that the substrate surface is both planar and at the endpoint elevation.